Home - on22s/Side-Story GitHub Wiki
Side Story
Formerly Alexandria Audiobook Nya.
Turn a book into a multi-voice audiobook, then listen, edit, and export it in your browser.
New here? Install the app → Make your first audiobook.
Side Story's core workflow has five tabs: Setup → Script → Voices → Editor → Result. Start with a short passage and a preset voice. You can add voice design, cloning, LoRA training, and dataset tools when you need them.
Start and configure
- Installation: recommended Pinokio setup, alternative installs, first-run downloads, and realistic platform/GPU expectations.
- LLM setup: connect a local or hosted model, use manual copy/paste transport, and understand which text leaves your machine.
- Your first audiobook: a complete short-book walkthrough, from upload to a checked download.
Make the book sound right
- Scripts and review: annotation, source fidelity, review, checkpoints, manual requests, and saved scripts.
- Voices: presets, cloning, designed voices, LoRA adapters, aliases, and reusable casts.
- Rendering and export: preview and fix chunks, run checks, render pending audio, and export MP3, M4B, chapters, or Audacity tracks.
Go further and keep your work safe
- Optional Voice Tools: Designer, Dataset, Training, Preparer, Voice Lab, and their dependency boundaries.
- Updates and backups: what to save before updating or resetting, including work that a saved script does not contain.
- Troubleshooting: connection failures, memory pressure, model downloads, broken audio, and useful bug reports.
- Acknowledgements: people and projects behind Side Story.
Two separate jobs: annotation and speech
The LLM turns source text into an editable script with speakers and delivery directions. Automatic annotation needs an OpenAI-compatible endpoint; manual transport lets you copy requests and paste responses instead.
The TTS engine turns that script into audio. Local Qwen3-TTS is included with the normal installation; its weights download when needed. An external Qwen3-TTS Gradio server is another option. Installing the app does not also install an LLM or all optional Voice Tools.
Before a full book
- Use books and voice recordings you have permission to process and distribute.
- Check where your configured LLM and TTS service run. Remote services receive the content needed for their tasks.
- Complete a short sample on your own hardware. Installer branches describe intended support, not a guarantee for every GPU, driver, or workload.
- Review the script against the source and listen to the audio before sharing it.
- Keep a backup of the active project and reusable voices. A saved script is not a complete audio-project backup.
About this guide
This wiki documents the Side Story application on main, verified against commit 8c7a33c on 10 October 2026. Guidance was checked against source and UI definitions; installation, GPU rendering, and platform performance were not tested for this documentation update. Newer app versions may change labels or behavior.